[m-rev.] Post commit review: Remove available-space field from region struct

Zoltan Somogyi zs at csse.unimelb.edu.au
Mon Dec 24 10:54:10 AEDT 2007


On 21-Dec-2007, Quan Phan <quan.phan at cs.kuleuven.be> wrote:
> Now I don't see why I decided to put MR_regionpage_allocated_size into
> MR_RegionPage. That field is to collect memory allocated into a region,
> so it should be better in MR_RegionHeader. I made this change.

Good; I agree.

> It would take no change to the current code if we use both macros
> with MR_REGION_PAGE_SPACE_SIZE = MR_REGION_PAGE_SIZE - 1.

That would be fine.

> You said to me
> before that the MMC has a limitation on the number of argument a functor
> can have. I think if it is reasonable we can derive the value of PAGE_SIZE
> from that number. 

Yes, that is a good point, and one I should have remembered earlier.
If I remember correctly, the maximum permissible arity is 1024, which means
that the smallest power-of-2 value of MR_REGION_PAGE_SIZE is 2048. Anything
less, and it would not have room for both a cell with a secondary tag and
1024 argument fields, and a link to another region page.

Zoltan.
--------------------------------------------------------------------------
mercury-reviews mailing list
Post messages to:       mercury-reviews at csse.unimelb.edu.au
Administrative Queries: owner-mercury-reviews at csse.unimelb.edu.au
Subscriptions:          mercury-reviews-request at csse.unimelb.edu.au
--------------------------------------------------------------------------



More information about the reviews mailing list